Michelin II - The treatment of rebates

Page created by Johnnie Maxwell
 
CONTINUE READING
Michelin II – The treatment of rebates*

                          Massimo Motta 1
             European University Institute, Florence, and
                Universitat Pompeu Fabra, Barcelona

                                   27 November 2006

*
  Forthcoming in a book on EU competition case studies, edited by Bruce Lyons (Cambridge U.P.).
1
 The author has not been involved in this case, and his information about the case was only and
exclusively drawn from public sources, such as the Commission Decision and the Court of First Instance’s
Judgment. I am very grateful to Chiara Fumagalli and Liliane Karlinger for their comments on a previous
draft.

                                                   1
1. Introduction

In 2001, the European Commission found that the French firm Michelin had – via its
various types of rebates - abused its dominant position in the French markets for new
replacement tyres and retreaded tyres for heavy vehicles, and imposed a fine of EUR
19.76 million to Michelin. 2 Two years later, the Court of First Instance upheld the
Commission’s Decision in its entirety. 3
In many respects, this case is exemplary of the strict formalistic approach followed in
abuse of dominance cases by the European Commission and the Community Courts,
which severely limit the possibility of dominant firms to resort to certain business
practices, such as exclusive dealing, rebates, tying. Indeed, the EU case law has so far
disregarded the actual effects of the allegedly abusive practices (the Commission does not
need to prove that exclusionary effects have indeed taken place, nor does it need to show
that consumers have been hurt), the mere possibility that they could distort competition
being enough for a finding of infringement of article 82 of the Treaty. As a matter of fact,
Michelin II is even stricter than previous decisions and judgments because for the first
time it is found that a dominant firm cannot even resort to pure (non-individualised)
quantity discounts, 4 a practice until then accepted by the Courts.
Perhaps, though, the very fact that it pushed a formalistic interpretation of article 82 to its
limit, will paradoxically and unintentionally make Michelin II a judgment which will
contribute in a positive way to European competition law. This judgment made it
apparent that in abusive cases economic, effects-based considerations are completely
absent, and contributed to open a debate on how to reform article 82 policy. 5
This chapter will be organised as follows. In Section 2 I will briefly describe the markets
at hand, and summarise the commercial policies followed by Michelin. In Section 3 I will

2
  Decision 2002/405 of 20 June 2001, published in the Official Journal of the European Communities on 31
May 2002.
3
  Case T-203/01, Manufacture Française des pneumatiques Michelin v. Commission, Judgment of the
Court of First Instance of 30 September 2003.
4
  Unless it can show that such discounts can be justified by per-transaction savings, see below for a
discussion.
5
  At the moment I am writing, the European Commission is still in the midst of this process of reform, and
– although it claims that it intends to undertake a more economics-based approach – it is not clear to what
extent this objective will be followed in practice.

                                                    2
report the main arguments followed by the European Commission and the Court of First
Instance. Section 4, which is the main section, will contain my own assessment of the
case: 6 in particular, I will criticise the fact that rebates may be found per se abusive,
without any need of formulating a coherent theory of foreclosure, without looking at any
evidence of the actual effects of the practices at issue, and without considering possible
pro-competitive effects.

    2. A short description of the case
In this Section, I will first briefly describe the relevant markets and then summarise the
commercial policies that made the object of the investigation.

The markets
The discussion of the product and geographic markets is important to understand the
case, but will be kept short here because it is not particularly controversial.
There exist several different types of tyres, and there is consensus among the parties that
tyres for trucks and buses are not substitutable with tyres for cars, vans, tractors and so
on. There is also a difference between new tyres which are sold in the original equipment
market (directly to truck and bus producers) and new tyres sold in the replacement market
(buyers are final consumers, which mainly purchase at specialised outlets). This case
concerns the latter market only..
A customer who needs to replace his truck or bus tyres may either buy new tyres or
retreaded tyres: 7 this is because the tread of a worn tyre casing can be renewed. There
exist two different processes for retreading tyres: mould-cure, prerogative of tyre
manufacturers, and pre-cure (by middle-size firms). Mould-cure and pre-cure retreading
are considered close substitutes.
The markets for new replacement tyres and that for retreaded tyres are the two relevant
product markets of this case. They are considered distinct because apparently the quality

6
  Michelin II, and the EU policy on rebates, are also discussed in two excellent articles, Spector (2005) and
Waelbroeck (2005).
7
  53% of the replacement market for truck tyres is composed of new replacement tyres, and 47% of
retreaded tyres.

                                                      3
of new tyres is perceived as superior (more reliable) to that of retreaded tyres, which are
cheaper, being sold at less than 50% of the price of new ‘brand’ tyres; but new ‘white
brand’ or ‘third-line’ tyres also more expensive than retreaded tyres. 8 The Commission
Decision does not put it in these terms, 9 but we know that in order to define relevant
markets we should use the available information to answer the SSNIP test question: could
a hypothetical monopolist of new replacement tyres profitably raise prices by 5-10%?
Only in this way could we appreciate the competitive constraint that the products exert to
each other. From what is reported in the Decision, we do not have data to answer this
question, but it is possible that the answer would be positive because of the higher
reliability of new tyres.
Finally, the geographic scope of these product markets is considered to be France, given
that the French market appears to have sufficiently distinct features from other markets
which are considered not to exercise a competitive constraint on the French market.

The industry
The industry for new tyres is quite concentrated both in the world and in the EU, the first
six producers accounting for around 75-85% of the markets. There are different leaders in
different markets: at the world level (according to 1997 data referring to all types of
tyres) Michelin and Bridgestone share the leadership with 18-20% of the market,
followed by Goodyear (15%), Continental (8%), Sumitomo/Dunlop (6%), and Pirelli
(5%). 10 In Europe, however (1995 data for all tyres) the leader is Michelin with 31% of
the market, followed by Continental (17%),Goodyear (13%), Pirelli (9%), Bridgestone
(8%), Sumitomo (8%). The relative competitive positions of the firms change further
when different EU countries are considered, Michelin being stronger in France, Pirelli in
Italy and so on.

8
   ‘Third-line brands’ are low quality tyres – often ‘white brands’ - produced in low-cost countries but
sometimes also by major manufacturers. They still have a limited penetration at the time of the Decision.
9
  The Commission still defines the relevant markets by looking at price similarities, physical features of the
products and so on. For a critique to this approach, see e.g. Motta (2004: chapter 3).
10
    Goodyear and Sumitomo merged in 1999, but the period of the infringement considered here was 1991 to
1998 included.

                                                      4
Public sources put Michelin’s market share for new tyres in 1995 at 55-60% (no data are
reported for the retreaded tyres market). Its leadership during the period 1991-1998 is
stable even though the market share is decreasing (see also below).
As for production capacities, Michelin has 11 of the 19 production plants in France, the
others being owned by Kléber (controlled by Michelin), Continental, and
Dunlop/Sumitomo with 2 plants each, and Bridgestone/Firestone, and Goodyear with 1
each.
Specialised distributors have 75-85% of the truck segment for tyres, their position of
strength being due to the fact that the choice, fitting and management of truck tyres
requires technical knowledge. Probably due to this, there is a growing tendency for
manufacturers to control distribution and vertically integrate.
Michelin owns the largest network among the 2225 specialised sales outlets (data refer to
all tyres), the chain Euromaster, which is composed of 330 outlets; 11 while
Sumitomo/Dunlop have 120 specialised outlets and Bridgestone/Firestone only 26.

As for the retreading market, Michelin does mainly mould-cure retreading, while its main
rival, Bandag, which operates through a network of franchisees, does pre-cured
retreading. Market shares for this market are not reported, although the Judgment
indicates that the pre-cure retreading segment is growing (75% in 2002). 12

Michelin’s commercial policy

Michelin’s commercial policy, central to this case, was composed of four elements. 1.
The general price conditions for professional dealers, consisting of a list price (‘invoicing
scale’) plus a system of rebates (quantity rebates, service bonus, progress bonus,
individual agreements); 2. the so-called ‘PRO agreement’; and 3. the ‘Club des amis
Michelin’, an agreement on business cooperation and assistance.

11
   Interestingly, Euromaster sells not only Michelin tyres, but also competing brands. We shall come back
to this point, which raises some doubts about the Michelin’s alleged intention to exclude rivals (why
selling competing brands, if Michelin intended to foreclose them?)
12
   The most common practice in France is ‘custom retreading’ (where the own tyres are retreaded), while in
the EU ‘standard-exchange retreading’ is more common.

                                                    5
In what follows, I briefly describe each of these elements.

1 – General price conditions (1980-1996)13
A dealer buys (new or retreaded) tyres according to the official ‘invoicing scale’, and
pays within 30 days of the end of accounting month. It then receives the total rebates as a
lump-sum only at the end of February of the following year.
Let us analyse the various types of rebates.

     i.   Quantity discounts
Two grids – one for new tyres, the other for retreaded tyres - showing the percentages
deductible from the list price for the entire turnover achieved represent the main quantity
discount scheme. As one can see from Tables 1 and 2 below, the larger the volume of
purchases of a customer the higher the rebate it can enjoy. For example, a buyer
purchasing 10,000 FF of new tyres in 1995 will receive a yearly rebate of 8.5%, i.e. 850
FF. Seen from a one-year window, this effectively amounts to reduce the unit price paid
by the buyer: if a tyre’s list price is, say, 100, the fact of having bought 100 tyres through
the year reduces the unit price of a tyre from 100 to 91.5. In other words, although it
takes the form of a lump-sum payment which takes place once a year, this rebate amounts
to a reduction of the unit price to be paid. If a buyer was able to anticipate correctly the
exact amount of tyres he will buy, and apart from the possible cash effects (the rebate is
obtained with a delay of some months), this scheme amounts to assigning a certain unit
price to any intended total volume purchase.
However, it should also be noted that in case of some additional unexpected purchases,
i.e., changes relative to the expected purchase volume, the effect at the margin is larger
than it may appear at first sight. Consider for instance a buyer who decides to order an
extra 10000 FF of tyres. This will bring the total amount purchased to 20000 FF, a
threshold subject to a 9% discount. However, the effective discount is larger, since due to
the new threshold met the buyer will also receive an extra .5% on the first 10000 FF

13
   Michelin changes its policy in 1997, after the entry into force of the French Law prohibiting selling at a
loss. Since the policy followed in 1997 and 1998 contains similar features as the previous one, for shortness
I do not describe it.

                                                     6
worth of tyres bought. Overall, therefore, the marginal purchase has a unit cost of 90.5
rather than 91. But the cost of an additional purchase may be even much smaller when a
very small number of additional units allow the firm to reach a higher quantity threshold.
For instance, if 9000 units command a discount of 7.5%, buying one additional unit
would bring the discount on all units to 8.5%: if the list price of one unit is 100FF, the
effective marginal cost of buying this extra unit would be 1.5 FF, which brings the actual
discount on this extra unit to 98.5%! 14

14
     100-(100x8.5%)-(9000x1%)= 1.5, where 100 is the unit price for the extra unit, 8.5% is the discount
applied to this extra unit, and 1% the additional rebate granted on all the 9000 units previously bought.

                                                       7
________________________________________________________________________
T/O 95         Rate    T/O 95         Rate    T/O 95          Rate   T/O 95            Rate
ii. Service bonus
The main objective of this type of rebate is to provide an incentive to the specialised
dealer ‘to improve his equipment and after-sales service’. To qualify, it was necessary a
minimum turnover with Michelin (the threshold passed from 160,000 FF in 1980 to
45,000 FF in 1996). The size of the bonus depends on the dealers’ compliance with
commitments they have taken in a number of areas; each commitment gives a number of
points; the more the points a dealer scores the higher the bonus, which is up to 1.5% until
1991, and up to 2.25% in 1992-96.
The criteria used to assign points include features related to the quality of the services
provided by the dealers, such carrying out staff training, having certain machinery and
know-how, and the level of quality of the facilities; features related to the dealer’s
behaviour with clients, such as supplying customers with new Michelin products,
promoting and advertising them; and providing Michelin with detailed information
(mileage of tyres, performance, statistics on sales, also comparing Michelin and
competitors) to help it plan its production. As for retreading, points are achieved by
providing roadside assistance for trucks; by showing knowledge about how to sort
casings; and by having Michelin tyres systematically retreaded by Michelin.

       iii. Progress bonus
This bonus rewards dealers who agree to commit in writing at the beginning of the year
to exceed a certain minimum base of purchases (which is fixed by common agreement
between the dealer and Michelin, and which depends on various factors), and manage to
exceed it.
The system changed over the years. In 1995-96, it specified, for sales between 30-999
tyres: 12% of the amount by which the base is exceeded, if the progress was lower than
20%; and 2% of the aggregate turnover if the progress was higher than 20%. For sales
above 1000 tyres: 15% of the amount by which the base was exceeded, if progress was
lower than 20%; and 2.5% of the aggregate turnover if progress was higher than 20%. If
the base was reached but not exceeded, a bonus of 0.5% of the turnover was granted.

       iv. Individual agreements

                                              9
This scheme was applicable to dealers who reached the maximum amount in the quantity
discounts table above (amounting to a turnover of 22m FF in 1995), who could then sign
a cooperation agreement with Michelin. The scheme required dealers to commit to
provide technical information and after-sales service; help launch Michelin products;
submit forecasts; get regular supplies from Michelin. The benefits foreseen were: an
extension of the discount tables (up to 2% additional rebate); a favourable system to
calculate the progress bonus; and an extension of the payment deadline (e.g., 60 rather
than 30 days).

2. The ‘PRO Agreement’
This scheme offers a dealer a rebate for each tyre given to Michelin for retreading. In
exchange, the dealer commits to a truck progress bonus, and will have Michelin retread
all the Michelin truck tyres which have reached the legal limit for the tread wear.
This scheme contains an element of tying, in that the maximum number of bonuses a
dealer can achieve is limited by the number of new Michelin tyres bought during the
previous year (and the bonus is conceived not as a cash payment but as a credit towards
buying new Michelin tyres).

3. The Michelin friends’ club
The “Club des amis Michelin” is an agreement that Michelin signed (bilaterally) with a
large number of sales outlets (375 in 1997, accounting for some 20% of the truck tyres
markets). According to the agreement, Michelin helps the club members to improve their
performance and professionalism; in return, members guarantee a certain level of
‘Michelin temperature’ (i.e., large enough volumes and market share).
More specifically, Michelin contributes to investment and training of the dealers; this
includes a financial contribution, if a specific target is attained; but also the transfer of
know-how in many areas; priority access to training courses (50% of courses at
Michelin’s training centre are reserved t members); transmission of data on market
trends; and exclusive distribution of BF Goodrich car tyres.

                                               10
Members’ commit to communicate balance sheets, and provide Michelin with
disaggregate sales statistics (but Michelin also carries out individual financial analyses,
and helps members to find solutions to their financial problems). They also permit
Michelin to carry out analyses of the outlets in areas such as staff qualification and
competence, quality of services and sales promotions, sales facilities; they promote
Michelin products by displaying advertising material, carrying product information, and
taking part in advertising campaigns; they commit ‘not to divert customer demand away
from Michelin’; carry enough stock of Michelin products so as to meet customer demand
immediately; have their first retreading made by Michelin; play an active role in truck
tyre sales and services.

3. The Commission’s findings
To prove that a firm has abused a dominant position, i.e. that it has infringed article 82 of
the Treaty, the Commission first has to find the firm is dominant in a given relevant
market, and then has to show the abusive nature of the firm’s practice.

Michelin’s dominance

In this case, the relevant markets are defined (see above) as the (i) new replacement truck
tyres in France and the (ii) retreaded truck tyres in France. The Commission finds that
Michelin has a stable leadership in these markets, with market share above 50%
(however, we know from the CFI judgment that Michelin’s share is decreasing over
time). Moreover, its strength is also due to its technological lead and expertise; its strong
reputation; the wide range of its products; and its strength in commercial and technical
services, demonstrated by its leading position in the French distribution channel (where
it owns directly sales outlets, and has strong links with a large number of other outlets
thanks to the Club des Amis Michelin).
These points of strength should therefore support the idea that Michelin is dominant, i.e.,
it has considerable market power: in other words, it can keep prices well above cost
because little demand would switch to rivals and importers, that according to the

                                             11
Commission are not competitive enough. Further, the Commission argues that buyers do
not have much power, as specialised dealers “must deal” with Michelin, else they would
lose credibility (since clients would ask for Michelin’s products).
The issue of dominance may deserve some further analysis (Michelin’s market share
appears to decrease over time, even if we do not have precise data on this; 15 further, the
Commission itself indicates at para. 212 that in the last decades, new firms have entered
the French market and have become more competitive), but the absence of data in the
Decision suggests that we should focus on other issues.
One element that I would like to point out for further discussion below, is that the
Commission stresses strong competition among dealers and indicates it as an element
which compels dealers to accept Michelin’s rebate schemes:

       “Competition between dealers [is] very keen and margins small (in 1995, for
      example, the average operating margin of French specialised dealers was 3.7% of
      their turnover). Under this pressure, independent dealers are constrained to obtain
      the best terms possible when purchasing Michelin products, especially as regards
      truck tyres. In order to preserve the competitive edge on which their survival
      probably depends, dealers do not hesitate to take part in the majority of
      ‘programmes’ and ‘agreements’ that they are offered, if as a result they can benefit
      from rebates or other additional economic advantages. Very slight variations in the
      rebate rates can prove to be essential to the dealers, who will do their best to make
      the most of them.” (Commission Decision, Para 206.)

The abusive nature of Michelin’s commercial policy

15
   At para. 176 of the Decision, the Commission reports some publicly available estimates which put
Michelin’s share in the new replacement tyres for trucks in France at 55% in 1995 and 51% in 1996. The
strongest competitors appear to be Dunlop (9% in both years) and Bridgestone/Firestone (whose share
drops from 12.5% in 1995 to 9% in 1996); Goodyear (6.5% in 1995 and 7% in 1996) and
Continental/Uniroyal (5% in 1995 and 6.5% in 1996) and Pirelli (3% in both years) follow, while minor
suppliers (including third line) grow from 9% to 14.5%.

                                                   12
According to the Commission, when faced by new entrants and stronger rivals, Michelin
has adopted a complex system of rebates and discounts to maintain dominance, thus
leading to the finding of Michelin’s infringement of article 82 and the imposition of a
large fine. In what follows, I review the Commission’s arguments, delaying my own
considerations to the next Section.

General price conditions
Quantity discounts are criticised by the Commission because the purchase of some
additional units may determine reaching of a higher quantity threshold and thus entitle
the buyer to a larger rebate on all units (additional and previously bought). Such rebates
(which indeed may determine a very low price associated with marginal purchases, as
discussed above) would therefore create a strong incentive to buy additional units from
Michelin, determining a loyalty-inducing effect.
The Commission also criticises, following an established case law, the fact that the
discounts would not reflect economies of scale. This is because the case law of the
Community Courts only allows discounts which are justified by transaction-specific
savings: if a firm can show that the unit cost of supplying in one particular transaction a
particular buyer with 100 units is lower than the cost of supplying the same buyer with,
say, 50 units, then a discount to that buyer would be justified. But claiming that a rebate
scheme would induce one or more buyers to purchase more over several transactions,
which in turn allows the firm to reduce unit costs of production, would be considered a
generic claim which does not qualify as efficiency gain.
The Commission also judges the scheme unfair, because: rebates are not paid until
February on the following year, which would determine cash flow problems for buyers,
and even oblige them to operate at a loss until the rebate is paid; dealers have to negotiate
the progress bonus with Michelin before they have received the quantity rebates for the
previous year, which would put them “in a weak psychological position during
negotiations” (para. 223); buyers would suffer from uncertainty, as until the end of the
year they do not know the actual cost of inputs.
I always find it difficult to evaluate fairness arguments in connection with competition
policy issues, but I would none the less like to stress the paternalistic approach of the

                                             13
Commission, which seems to view buyers (which in this case are firms, not final
consumers) as being unable to make forecasts and organise their business.
Finally, according to the Commission this system would have market-partitioning effects,
because rebates are applied only to purchases from Michelin France, thereby making
parallel imports difficult, as tyres bought outside France do not qualify for the thresholds
established by the rebate schemes. (Recall that preventing parallel imports, that is
arbitrage between different EU member states, is a very serious violation of EU
competition law.)
Curiously, it is in this section devoted to parallel imports, that the Commission goes the
closest to formulating a theory of foreclosure in the Decision:

      “…thanks to its market shares, Michelin was able to absorb the cost of these
      rebates, while its competitors were unable to do likewise and therefore had to either
      accept a lower level of profitability or give up the idea of increasing their sales
      volume.”(Commission Decision, para. 241)

The service bonus scheme shows according to the Commission similar abusive features
as quantity discounts. In particular, it is unfair, because: a) it is subjective (Michelin has a
wide margin of discretion in assigning ‘points’ to buyers); b) it asks dealers to provide
market information, which would not be in their interest and for which they would have
no return (e.g., in form of market studies); c) subjectivity is inevitably a source of
discrimination (recall that a dominant firm may not discriminate among customers,
discrimination being abusive). It is also loyalty-inducing, in that reaching a minimum
turnover is necessary for a dealer to qualify and strengthen links with Michelin; and
because it assigns points if Michelin tyres are systematically given to Michelin for
retreading, which effectively amounts to a tying practice.

As to the progress bonus, (as well as the ‘achieved target bonus’ which replaces it in
1997) it is particularly abusive because it is loyalty-inducing, by pushing dealers to buy
more than previous years (or meet the target), thus denying sales to rivals; and unfair,
because: a) it is discriminatory (two dealers who buy the same quantity but have different

                                              14
bases get different rebates); b) it “creates insecurity” in dealers (they do not know if they
will get the bonus).

Individual agreements are equally abusive, because they give extra incentives to large
dealers, and put pressure on them to buy only from Michelin, so that they can reach the
highest rebates.

The other rebate schemes

Similar arguments as to those made above are also used by the Commission to explain
why the remaining rebate programmes are abusive.
The ‘PRO agreement’ contains a ‘double-tying’ component, since Michelin casings have
to be sent exclusively to Michelin for retreading. The effect of this practice is abusive –
according to the Commission - because Michelin would use dominance on the new tyres
market to strengthen it in the retreading market and vice versa. Further, the fact that the
bonus is limited to the number of new Michelin tyres sold during the previous year (no
matter how many tyres are sent for retreading) would imply that dealers are discouraged
from buying rivals’ new tyres, for the following year the bonus would be lower.
As for the Michelin’s Friends Club, it would also contain abusive obligations, in that
dealers: i. need to achieve a certain ‘temperature’ (i.e., proportion of sales from
Michelin); ii. cannot divert spontaneous customer demand away from Michelin; iii. need
to carry a sufficient stock of Michelin products to meet demand immediately. These
obligations would aim at eliminating competition and maintaining dominance, with an
effect similar to a ‘fidelity clause’, further aggravated by the fact that Michelin has also
an integrated network which ensures it a large market share. Finally, the Commission
also objects to the ‘excessive’ monitoring of dealers’ business, which Michelin would use
to control dealers and make sure they do not buy from its competitors.

There is no doubt that Michelin’s rebate programme contains some elements of
aggressive competition, as we shall discuss below. For the time being, it is important to
underline that the Commission’s arguments are either non-economic in nature (the

                                              15
alleged unfairness of the practices followed by Michelin), or are insufficiently motivated.
As we shall discuss below, it is conceivable that rebates (as tying and exclusivity
provisions) may exclude rivals, but here there is neither an attempt to explain why
foreclosure should occur, nor a presentation of evidence which indicates that foreclosure
was indeed taking place, not to mention the complete absence of possible pro-competitive
justifications for the rebates or of an assessment of the effect of such practices upon
consumers.

The Commission’s Decision is appealed by Michelin, but the Court of First Instance
upholds the Decision, and accepts all the Commission’s arguments (which in many cases
just followed an established case-law).
The main element of novelty in the Judgment is the fact that for the first time the Court
rejects the use of pure quantity discounts. According to the Court, they induce loyalty
because the rebates were calculated on the overall turnover, and are abusive because not
justified by cost savings:
     “…a rebate system in which the rate of the discount increases according to the
     volume purchased will not infringe Article 82 EC unless the criteria and rules for
     granting the rebate reveal that the system is not based on an economically justified
     countervailing advantage but tends, following the example of loyalty and target
     rebate, to prevent customers from obtaining their supplies from competitors (…)”
     (para. 59 of the Judgment, italics added )
      “A quantity rebate system has no loyalty-inducing effect if discounts are granted on
     invoice according to the size of the order. If a discount is granted for purchases
     made during a reference period, the loyalty-inducing effect is less significant where
     the additional discount applies only to the quantities exceeding a certain threshold
     than where the discount applies to total turnover achieved during the reference
     period.” (Para. 85 of the Judgment)

But perhaps the most striking, and potentially important, element in the Judgment is the
way in which the CFI dismisses Michelin’s argument that the rebate system did not have
any effect on competition, as M’s market shares and prices were falling:

                                             16
“[…] In any event, it is very probable that the falling in the applicant’s market
       shares (recital 336 of the contested decision) and in its sales prices (recital 337 of
       the contested decision) would have been greater if the practices criticised in the
       contested decision had not been applied.” (para. 245 of the Judgment)

As one can immediately see, the Court’s arguments may completely undermine an
effects-based approach: for any evidence that a firm may produce in support of its claim
that rivals have not been hurt, and/or that consumers have benefited from the practice, the
Court could always counter such evidence with an argument (which does not need to be
supported by evidence) that rivals and consumers would have benefited even more had
the dominant firm not engaged in the (abusive) business practices at hand.

An economist’s assessment of Michelin II
Most business practices which are potentially exclusionary – and this holds not only for
rebates, but also tying, exclusive dealing, refusal to supply - may also have pro-
competitive effects. Since the objective of competition policy is not to protect
competitors, but to protect competition and increase welfare, economic analysis suggests
to undertake the following four-step approach in order to find whether a firm has engaged
into abusive practices: first, find whether the firm is dominant, that is if it has
considerable market power; second, identify whether the practice has indeed possible
anti-competitive effects, including the formulation of a coherent hypothesis about the
strategy of the firm; third, analyse the possible pro-competitive (efficiency) effects of the
practice at hand; fourth, balance the anti- and pro-competitive effects, that is, carry out an
assessment of the net effects on consumer (or total) welfare.16

16
  See for instance Motta, 2004. The particular way in which the test is proposed does not matter much. The
important thing is that an explicit analysis of both the anti-competitive and the pro-competitive effects, and
an estimation of their net effect on welfare, is carried out. Gual et al. (2005) suggest that there is no need to
separately show that the alleged violator has market power, but this is in my opinion a useful screening
device, which has the further advantage of bringing the test in line with EU competition law.

                                                       17
The Commission and the Community Courts, instead, basically adopt a per se rule of
prohibition of rebates (but also exclusive dealing and tying) by a dominant firm. They do
not present a theory of why the dominant firm may use rebates so as to foreclose rivals,
do not consider necessary to show any actual foreclosure, do not analyse the possible
efficiency rationale behind the practices, and do not try to assess whether ultimately the
practices would have harmed or benefited consumers. 17

In what follows, I try to assess the Michelin II case from the perspective of the economic
approach delineated above. However, I do not discuss the issue of dominance (which is
simply assumed to be proved) ad focus instead on anti-competitive effects and the
possible pro-competitive ones.

Do rebates have anti-competitive effects?
Rebates can be defined as discounts applicable where a customer exceeds a specified
target for sales in a defined period. There are different types of rebates, according to the
sales ‘target’ given to the customer. Rebates can be conditional (i) on increasing
purchases; (ii) on buying only (or in a sufficiently high percentage) from the given
supplier; or (iii) on buying over a given amount of units (quantity discounts). They can
also be individualised or standardised.
Interestingly, Michelin’s commercial policy included more or less the whole range of
possible rebates. The progress bonus belongs to type (i), as it established a minimum
turnover, which represented the base to measure ‘progress’, that is the extent to which the
target was met or exceeded, in turn determining the discount. The Michelin’s friends club
was of type (ii), since it required a ‘certain temperature’, i.e., a sufficiently high volume
and share of purchases from Michelin to qualify; the service bonus and the individual
agreements also contained elements of type (ii) by requiring a certain minimum turnover
to qualify (the scheme does not apply if sales are below the threshold); as for the quantity

17
     In the US, there is a very different approach, which considers rebates as “competition on the merit”, and
puts the burden to prove anti-competitive behaviour on the plaintiff. However, recent decisions such as
LePage and Dentsply may signal a change. See Kobayashi (2006) for a discussion of the US policy on
rebates.

                                                       18
discounts, they belong to type (iii), since the different discount rates were conditional on
the purchase of different number of units (for any given amount bought during one year,
one could easily compute the unit cost associated to it). Further, the rebates offered by
Michelin can be either individualized (indeed, the Commission and the Court criticize the
fact that both the service bonus and the progress bonus are discretionary and
discriminatory, since two identical buyers can be offered different rebates), or
standardized (quantity discounts are in principle identical to any costumer).

Conceptually, the first two types of rebates are similar to exclusive dealing provisions, in
that they try to induce a buyer to purchase all (or most of) its inputs from the same
supplier, either directly (specifying that a certain percentage or volume is purchased) or
indirectly (by conditioning the discount to buying the same or increased number of units
than in the previous year). 18
In this perspective, we may rely on the literature on exclusive dealing to understand why
rebates may foreclose rivals. 19 In particular, Rasmusen et al (1990) and Segal and
Whinston (2000) have showed that if buyers cannot coordinate their purchase decisions
and a new firm needs to secure a certain minimum number of buyers to operate
efficiently, exclusive dealing allows an incumbent monopolist to exclude the rival. 20
Indeed, the incumbent monopolist just needs to induce a certain number of buyers to
purchase exclusively from it, thereby preventing the entrant from achieving the minimum
number of orders necessary to operate profitably, and effectively obliging all remaining
buyers to buy from the incumbent as well.
Although there are some differences between rebates and exclusive provisions (unlike the
latter, the former practice does not include any ex ante commitment on the side of the
buyer), one could presumably formalize a story whereby an incumbent could use
discounts conditional on buying more than in the past, or on buying at least a certain
percentage of input needs, to prevent rivals from reaching a minimum size of its business.
18
   Indeed, such types of rebates have always been considered ‘abusive’ in the EU because they are said to
amount to exclusive dealing, a practice which is not tolerated by dominant firms.
19
   See Motta (2004: Section 6.4) for a presentation of the literature.
20
   See also Bernheim and Whinston (1998). In their model, in which markets appear sequentially, an
incumbent is able to use either explicit exclusive dealing provisions or a sort of quantity discount to
exclude an entrant from the first market; since the entrant needs to sell in both the first and the second
market to operate profitably, foreclosure will arise.

                                                    19
As for quantity discounts, it is not straightforward to see at first sight their exclusionary
potential: how can an incumbent use non-individualised quantity discounts so as to
exclude a more efficient firm? Note that – unlike exclusive dealing – under rebates there
is no contractual commitment for the buyer to buy only from the incumbent: here the
incumbent can offer only a price-quantity menu which can in principle be matched by the
rival.
Karlinger and Motta (2006) take seriously the argument of the Commission and the Court
of First Instance and show that it is indeed possible for an incumbent to exclude a more
efficient rival, even if the latter can use the same rebate scheme.
In their model with buyers of differing sizes, the incumbent has an incumbency
advantage given by an established customer base that the (more efficient) rival firm does
not have. In order to be profitable, the rival needs to attract a certain minimum turnover,
while the incumbent has already reached it. 21 In this model exclusion may arise even
under linear pricing (because of miscoordination of buyers – if a buyer expects that all
other buyers would buy from the incumbent, s/he would have no incentive to buy from
the entrant, since its orders would not be sufficient to trigger entry). However, the authors
show that quantity discounts have a higher exclusionary potential. To understand why,
consider first the benchmark situation where the incumbent could make explicit price
discrimination, and suppose that the entrant is not able to reach the minimum threshold
size if it does not sell to the large buyers. In this case, the incumbent could break an
equilibrium where the rival sells to all buyers by making a below-cost price offer to the
large buyers, while recovering profits by imposing the monopoly price to the small
buyers (who, failing the rival firm to achieve minimum size, would have no other
possibility than to buy from the incumbent). Karlinger and Motta (2006) show that –
unless the incumbent is considerably less efficient than the entrant – under price
discrimination the only type of equilibrium arising in the game is the exclusionary one,
where only the incumbent serves.

21
  The main model in Karlinger and Motta (2006) considers an industry with network externalities, and
assumes that the entrant needs to achieve a sufficiently large number of buyers for them to derive utility
from the consumption of the network product. Here, I am describing the mechanism of the paper assuming
economies of scale in production rather than in demand, but the two are equivalent.

                                                    20
Next, observe that the incumbent can use this ‘divide and rule’ strategy even through
quantity discounts, which are nothing else than a form of implicit price discrimination.
Indeed, the only difference between a scheme which explicitly offers different prices to a
small and a large buyer and a quantity discount scheme is that under the latter buyers
have to self-select, i.e. can choose freely which offer to accept. However, provided that
the price offers are not ‘too imbalanced’, so that a small buyer does not want to behave as
if s/he were a large buyer, some price discrimination can still be achieved, thus
determining exclusion, at least under some conditions.
Karlinger and Motta show that the more aggressive the pricing scheme (linear pricing is
the least aggressive, since the firm cannot discriminate, followed by quantity discounts
where in order to discriminate successfully the self-selection constraint of buyers must be
satisfied, and finally by explicit price discrimination) the more exclusionary potential it
has, that is the more likely that at equilibrium the incumbent operates in monopolistic
conditions.
However, they also show that when comparing equilibria where exclusion cannot be
achieved (for instance, if the incumbent is much more inefficient than the rival, or if the
minimum threshold size needed by the entrant is small), the welfare ranking is exactly the
opposite: the more aggressive the pricing scheme, the lower the prices at equilibrium, and
therefore the higher the welfare outcome.
These results demonstrate the fundamental dilemma that one faces when establishing the
policy towards rebates (and pricing schemes in general): by prohibiting dominant firms
from using aggressive pricing policies (such as rebates), one lowers the risk that
exclusion will occur; however, this would at the same time prevent competition from
taking place, which will promote entry, but also higher prices.
Facing such a trade-off, different policy options are possible. (i) Always allowing rebates
(or other ‘aggressive’ pricing policies), for instance by putting the burden of proving anti-
competitive behaviour on the plaintiff and imposing a very high standard of proof (which,
by and the large, has been the US case-law so far, at least until the LePage and
Dentsply’s District Court decisions). (ii) Imposing a per se prohibition of rebates by
dominant firms, which by and the large corresponds to the current EU policy. (iii) Use a
rule of reason approach, which admits that rebates may be exclusionary, but also

                                             21
recognise that they can also be an instrument of market competition and determine
important welfare gains for consumers. The third approach is certainly more difficult to
implement because it calls for detailed investigations, but it is the only one which is
consistent with economic analysis. If one decides to follow this route, one needs to carry
out a detailed analysis of the case and understand to which extent in the case at hand it is
sufficiently likely (rather than simply conceivable) that rebates might achieve exclusion
of rivals.

Let us now come back to our case. We know that theory admits the possibility that the
various types of rebates used by Michelin could be used for exclusionary purposes. 22 But
how likely is it that in the facts of this particular case, rebates would have the effects of
foreclosing rivals?
In all the models mentioned above, rebates (and exclusive dealing) are anti-competitive if
they exclude competitors. But is it credible that Michelin’s competitors would exit the
market, or at least be relegated to small niche markets? From the description of the
industry, we know that Michelin is certainly the leader of the French market, but its rivals
are no minnows, and even in France have strong positions (also in distribution). Further,
it does not seem that Michelin’s policies – even though they span over a long period of
years - 23 have had such strong effects on its rivals, whose collective market share – as
we are told in the CFI Judgment – is increasing over the years.
Perhaps we could be more convinced of Michelin’s anticompetitive strategies if it had
been showed that it was selling below cost to some groups of customers (as we have seen
above, discriminatory strategies might help a dominant firm to exclude by denying key
buyers to the rivals) but there is no information about this point in the case.
Furthermore, we know from the literature on exclusive dealing that when there is strong
competition among buyers (a fact that is emphasised by the Commission), it is less likely,
rather than more likely as the Commission claims, that the incumbent will manage to

22
   Apart from the similarities to exclusive dealing, we should also mention that the PRO agreement
contained some features which make it similar to tying, and that tying may be used to exclude (see e.g.
Whinston, 1990).
23
   Para. 359 of the Commission Decision recites: “The infringement extended over a period of 19 years or
more, since the commercial policy at issue was in operation at least from 1980 onward, and…Michelin
agreed to amend its agreements with effect from 1 January 1999.”

                                                   22
exclude an efficient firm. Indeed, Fumagalli and Motta (2006) show that if buyers
compete fiercely in the downstream retailer market, each of them will have a strong
incentive to deviate from the incumbent and buy from a more efficient supplier, in order
to guarantee itself a competitive advantage over the other retailer-buyers (and if they steal
business from the latter, they manage to (endogenously) increase the size of their orders,
allowing the entrant to reach its minimum threshold size).
Finally, a natural question which arises from the reading of the proceedings is the
following: if Michelin really intended to exclude rivals, why did it sell rivals’ products in
its Euromaster outlets?

Possible efficiency effects of rebates
Neither the Commission nor the Court of First Instance consider the possible pro-
competitive effects that Michelin’s rebate schemes may have (with the exception, as
indicated above, of transaction-specific cost-savings), whereas economics does suggest
that rebates may have efficiency effects.
Let us continue for convenience to divide rebates in two categories: those that ‘resemble’
exclusive dealing provisions and those which consist of quantity discounts. For the first
category, one should recall that exclusive dealing may promote investments. 24 In
particular, Segal and Whinston (2000a) and Motta and Rønde (2006) show that exclusive
dealing is welfare beneficial if it helps protect investments made by the supplier which
have the effect of improving the retailer’s productivity in both selling the supplier’s
products and the other suppliers’ products. In the case at hand, rebates may serve the
purpose of trying to mimic exclusivity provisions (recall that exclusive dealing by a
dominant firm is not allowed in the EU), and reduce the risk that an investment made by
Michelin into a retailer’s premises or human capital is then used by the retailer to sell
products produced by Michelin’s rivals. Although we do not have an explicit discussion
of such efficiency effects in either the Decision or the Judgment, there are some indicia
that rebates may have served the purpose of protecting Michelin’s investments. Several
clauses in the rebates schemes (for instance in the Service bonus and in the Club des Amis

24
     See Segal and Whinston (2000b) and references cited therein.

                                                     23
Michelin programmes) explicitly aimed at improving promotion, advertising and services
provided by dealers; and in some of these schemes, Michelin committed to provide
training, investments and funds. It seems logical that it does not want to invest in a dealer
who sells mostly rivals’ products. And in turn, by requiring a retailer to buy mostly from
it, Michelin would invest more into this relationship, which would be welfare-beneficial.

As for the other category of rebates, quantity discounts, we have already seen above that
they represent an instrument of price competition, and therefore – to the extent that they
do not exclude rivals - are inherently beneficial to consumers.
A slightly different way to look at quantity discounts is that they can be seen as a form of
two-part tariff: the more a buyer purchases, the lower the unit price paid. And we know
that this is going to reduce inefficiencies: when dealers’ variable cost decreases, so will
final consumers’ prices, causing an increase in welfare.
Furthermore, quantity discounts may indeed be justified by scale economies (even if not
for any single client and transaction in isolation, as the CFI requires): the more a supplier
produces and sells, the lower its production costs.

Although we do not have enough information in the published case material to quantify
the importance of such efficiency effects, there seems to be enough ground to conclude
that the efficiency effects of rebates should have been considered seriously.

Conclusions
The European Commission and the Community Courts de facto consider rebates by a
dominant firm per se abusive, as Michelin II demonstrates. But economics teaches us that
rebates may serve as a way to foster investments into the relationship between a seller
and its retailers; furthermore, like any reduction in prices and to the extent that they do
not exclude more efficient rivals, they have beneficial welfare effects. This should call
for a completely different approach in dealing with rebates, where first the exclusionary
potential of rebates is analysed, and then possible pro-competitive explanations are
considered, and if appropriate weighed against the anti-competitive ones.

                                             24
In this case, the Commission and the Court have made no effort in trying to formulate a
coherent theory of why Michelin’s rebate programmes had anti-competitive effects.
Likewise, they have not considered possible efficiency effects arising from such
programmes, while it is conceivable that by reducing prices in the market and by giving
incentives to invest more, considerable pro-competitive effects may have arisen from
them.
From the information available, instead, it does not seem particularly likely that
Michelin’s rebates could exclude (efficient) rivals. In fact, Michelin’s market position has
worsened over time despite it had used such rebate schemes. Nor can one exclude that
these schemes could achieve important efficiency gains which also benefited final
consumers. Had the Commission and the Court followed an economic approach, rather
than a form-based one, they would have probably concluded that Michelin’s commercial
policy was not abusive.

References
Bernheim B. Douglas. and Michael D. Whinston. 1998. "Exclusive Dealing." Journal of
     Political Economy. 106: 64--103.
Fumagalli, Chiara and Massimo Motta (2006). •"Exclusive dealing and entry, when
     buyers compete", American Economic Review, 96(3), June 2006, 785-795.
Gual, Jordi, Martin Hellwig, Anne Perrot, Michele Polo, Patrick Rey, Klaus Schmidt, and
     Rune Stenbacka.(2006). “An Economic Approach to Article 82”. Competition
     Policy International, 2(1): 111-154.
Karlinger and Motta (2006). •"Exclusionary Pricing and Rebates in a Network Industry",
     EUI, mimeo.
Kobayashi, Bruce H. (2005). “The Economics of Loyalty Discounts and Antitrust Law in
     the United States”, Competition Policy International, 1(2): 115-148.

                                            25
Motta, Massimo (2004). Competition Policy. Theory and Practice. Cambridge University
     Press.
Motta and Rønde (2006). "Exclusive contracts: between foreclosure and protection of
     investments", EUI, mimeo.
Rasmusen, E.B., J.M. Ramseyer and J.S. Wiley Jr. 1991. "Naked Exclusion." American
     Economic Review. 81: 1137-1145.
Segal, Ilya.R. and Michael D. Whinston. 2000a. "Naked Exclusion: Comment."
     American Economic Review. 90: 296--309.
Segal, Ilya R. and Michael D. Whinston. 2000b. "Exclusive Contracts and Protection of
     Investment" Rand Journal of Economics. 31: 603-633.
Spector, David (2005) “Loyalty Rebates: An Assessment of Competition Concerns and a
     Proposed Structured Rule of Reason”. Competition Policy International, 1(2), 89-
     114.
Waelbroeck, Denis (2005). “Michelin II: A per se Rule Against Rebates by Dominant
     Companies?”. Journal of Competition Law and Economics, 1(1), 149-171.
Whinston, Michael D. 1990. "Tying, Foreclosure, and Exclusion." American Economic
     Review, 80: 837--859.

                                         26
27
You can also read